Rebecca Adisa

Rebecca AdisaJul 6, 2026

I asked an AI to audit its own work. It did not go the way I expected. I used Figma Agent to generate four screens for a habit tracker app and then turned the agent on its own work, asking it to review every screen from five different user perspectives. A first-time user, an elderly user, someone with ADHD, someone using one hand, and someone with poor vision. For each role, it identified what the user was trying to do, where they would get confused, and why. It created a structured priority fix list, and when I asked it to apply the changes, it regenerated all four screens with the fixes already implemented. The feedback was more detailed than I expected. It went beyond surface-level observations and connected each issue back to a specific type of user and why they would struggle. But here is the thing. Every insight came from generic design knowledge. Rules and best practices it was trained on. It cannot factor in your business decisions, your UX research, or why a particular design choice was made in the first place. Those judgment calls still require a human designer. What has your experience been using AI to audit your own design work?

Rebecca AdisaJun 22, 2026

AI built an app. It still needed a designer to fix it. That designer is me. The AI and software team at Alluvium had been building internal apps using AI tools, moving fast and shipping quickly. During the presentation, stakeholders noted that the applications needed improvement, especially in their visual design and the consistency of the user experience across screens. So they called me in to help address the challenge. I will admit there is a bit of irony here. I have spent the last few months leaning into AI, learning new tools and how design and AI can work together. And now here I am, starting Monday, auditing AI-built apps because they could not stand on their own without a human eye. But honestly, that is exactly the point. AI can write code. AI can generate layouts. AI can move fast. What it still cannot do is think the way a designer thinks. It does not understand visual hierarchy the way a trained eye does. It does not catch the small inconsistencies that quietly break trust in a product. It does not ask why a user would feel confused on a particular screen or whether a flow actually makes sense from a human perspective. That is the gap I am stepping into this week. I will be reviewing these apps, looking at usability, consistency, and overall experience, and turning AI-generated output into something that actually feels easy to use. If anything, this is proof of what I have believed all along. AI is an incredible tool, but it still needs a designer in the room. Have you had to clean up or review something AI-generated recently? I would love to hear how it went.

Rebecca AdisaJun 25, 2026

The best features are not always the ones users ask for. Sometimes they are the ones that make users wonder how they ever managed without them. Church admins send a lot of messages. Birthday messages, event reminders, service announcements, and follow-ups for first-time visitors. Before we added the Scheduled SMS feature, every single one of those messages had to be sent manually and in real time. Which meant, if you forgot, it did not go out. And if you were busy, it did not go out. The team anticipated this problem before users had to live with it long enough to complain about it. The new Scheduled SMS screen gives admins a single place to see every message they have lined up, who it is going to, how many units it will cost, and exactly when it is going out. And if anything changes, a three-dot menu on each row lets them send immediately, modify the schedule, edit the message, or delete it entirely without leaving the screen. For a church admin managing hundreds or thousands of members across multiple schedules, this is not just a convenient feature. It is the difference between a communication strategy that runs itself and one that depends entirely on someone remembering to press send. That is the kind of problem good design solves before the user even knows they have it.